Hello and ahoy to another edition of AI versus Marshy!

It’s been a challenging few weeks - so I’m pleased to get this newsletter to you at a reasonable time.

This newsletter charts AI’s trajectory curve and our 1 year anniversary is next year - if you like the newsletter can you forward it someone you think might be interested and ask them to visit:

​https://www.lukemarshall.net/newsletter ​

So they can subscribe themselves.

Thanks! 🙏

This week covers:

• The best is yet to come - a novel look at AI’s utility
• Reviewing Audiopen - a voice note taker I’m paying for
• How to write an AI newsletter - I’ll walk you through my workflow as its a common question

Lots to share with you this week, so let’s get the foul, make the shot, and go for the 3-point play (Go Knicks! 🏀)

-Marshy

A novel look at AI’s utility

​Via Fraser @ Entropy and The AI Crowd is Mad.

Two good articles looking at the state of AI from a few months ago.

Tim and Fraser both disaggregate hype from reality and point to where the hype focus has been on AI versus where the reality and value lies.

Here’s some takeaways for you, followed by some comments:

• The discussion around the capabilities of LLMs (see: chat bots) needs more nuance
• There’s no doubt there’s been investor hype
• Take in media (podcasts) and social media (gross) pare down realities and over-simplify things
• We really don’t understand what’s really possible with LLMs yet as we’re still figuring it out

I’ve nerded out on charts.

If you’re an older reader you would have seen me reference the hype adoption cycle.

And if you’ve onboarded recently you might remember the tech mass-market adoption curve (hint: AI is moving faster than mobile phones).

Well Fraser shares another chart:

​

The argument here is that we’re nowhere near peak explosion and it’s yet to come.

And when it does it will fast.

I regularly try and take myself outside my own bubble.

Recently I talked about what’s coming with my mother.

She’s a career nurse and clearly doesn’t follow tech like I do.

I spoke to her about ChatGPT and gave her demo with voice.

It started with:

And got a generic answer.

Then I asked something relevant for Mum (prompted by her).

The answer was more impressive to her:

Absolutely, Marshy! Sewing a torn jumper is a pretty straightforward task. Here's a simple way to do it:

1. Start by turning the jumper inside out. This way, the repair will be less visible from the outside.
2. Thread a needle with a thread color that matches your jumper as closely as possible. If you're working with a thick fabric, consider using a stronger thread.
3. Begin near the end of the tear, making a small knot in the thread to secure it.
4. Use a stitch that suits the fabric and the nature of the tear. For many fabrics, a basic running stitch or a more invisible ladder stitch works well.
5. Carefully sew along the tear, making sure the stitches are even and pulling the fabric together without puckering.
6. Once you've reached the end of the tear, secure the thread with another knot.
7. Trim any excess thread and turn the jumper right side out to check your work.

It might take a bit of practice to get your stitches neat and almost invisible, but it's a handy skill to have! How does that seem for a start?

That’s Mum’s language.

It still wasn’t impressive enough for her to consider downloading the app or paying for a premium subscription - the reality is we’re not there yet.

My friend linked me on this take on a podcast he listens to (don’t send me podcasts I have enough to sift through!)

One way Meta (Facebook, IG, WhatsApp etc) could win is by making AI accessible to people like my Mum.

How I write an AI newsletter (and keep up)

Via me.

It’s a question I get a lot - so I thought I’d walk you through my workflow for creating this newsletter.

Ironically - I don’t use a lot of AI tools for this, although I’m aware of many ways I could summarise email digests and scan for keywords.

I’ve enjoyed going into the weeds and uncovering things to share with you.

Let’s go!

Throughout the week

• While browsing or reading emails, I save articles that catch my interest, either personally or for the newsletter
• On the web, I click the Reader Chrome extension; on my phone, I tap a button to Add to Reader; and for interesting emails, I forward them to a special email address that sends them to Reader. This has become habitual over time.
• During the week or the day before writing day, I’ll skim through my Marketing and Newsletter category in Gmail. Sometimes it piles up, but it’s a pretty good system.
  • All emails with “Unsubscribe” go in there
  • I have Keyboard shortcuts turned “on” and cycle through them quickly with label:”Marketing and Newsletter” label:unread
  • I filter through fairly quickly, and will save the relevant notes into “Newsletter notes”

​

Before writing

• I use a simple text editor called Typora that supports Markdown and start a new document called “Newsletter notes”
• I go through Reader - it’s a tool that lets me save documents, highlights from my Kindle, PDFs, articles, and more.
• Articles that are good fodder for an article go into “Newsletter notes” and I’ll add supporting article links where it’s needed
• I’ll also make sure I’ve got enough visuals for the newsletter, saving images, generating new ones, and labelling them

Writing time

• This part is really straightforward, I’ll open up a new text document called “Newsletter” and punch my words out in that
• Drafting usually takes 30-45 minutes and usually hits around 800-1200 words
• It’s worth adding that it used to take much longer

Publishing

• I use Convertkit to manage my list and publish - it’s more expensive than entry-level tools, but not insane cost-wise
• I can copy + paste the newsletter in, and then I manually add in the images 1-by-1 - I’m sure there’s a better way and this is the most painful part
• I’ll proof the email, and then send a preview email to one of my addresses - when I don’t do this I’ve been caught out 2-3 times and it hurts!
• I’ll then schedule it for 15 minutes later so if there’s a last-minute mistake I still have a chance to reverse it

Post-sending

• I’ll do my video wrap-up version of the newsletter a day later and upload it to LinkedIn and YouTube​
• It’s been a fun addition and I’ve definitely got better at presenting to camera over time

This is a long-time play for me and ladders into other things I do but in a complementary way. I had a habit of losing interest in things over time and knew I wanted to slay the dragon. I gave myself the goal of getting to 100 with the goal of getting better and it’s working.

My numbers aren’t huge and it costs me around 5-10 hours per week. I do know its worthwhile however and have had business opportunities and co-founder opportunities emerge off the back of this interest.

I also know from my experience as a marketer that email is where it’s at, and that AI is a long-term game-changer.

So doing a newsletter on AI that’s authentic is a long-term winner.

If you have any questions - please shout!

Happy to help my readers :)

Audiopen - a note taking app I am here for

​Via Audiopen.ai.

I used to review and play a lot with new AI tools. Then it began to feel gimmicky.

I’ve been using Audiopen for about a month now and am happily paying for the premium version.

Let me share why:

• Talking out my thoughts is super helpful for someone with ADHD - I can word-vomit and in doing so it helps declutter thoughts and get things moving
• I’ve been doing this since I got my Hyundai Kona in 2021. It has a voice recorder function that lets me go on a rant, and sometimes I’ll listen back to it and hear what I need to hear
• I took this a step further when I got my trucking headset. It’s ugly, but I can go for a walk and talk through my thoughts (as well as do phone calls like a normie person) and it’s super clear. The first time you walk past someone on the street while talking to yourself is crazy paralysing - it’s worth the discomfort though as it’s helped my workflow immeasurably
• I’ve tried using ChatGPT for this purpose but it sucks. It doesn’t understand pauses, will interrupt and jump in to be helpful. It will even do it when I specifically say - “you’re not to say anything until I get all of this out and say the codeword: project barracuda” - it will still chime in 😤

Enter Audiopen.

It doesn’t try and do too much and lets you narrate what you’re saying and it will translate it back to you in clarified language.

You then get to choose the way it translates things - in legalese (for some weird reason), casual language, bullet points, and more.

Skip the legalese:

​

But the casual language helped me write the previous article while driving:

I’ve been using it to plan and talk through my day, and for talking through something I’m stuck on.

My main gripe with it so far is that on mobile it runs with a web app - which will time out alongside when you’re phone is set to.

I gave the feedback and the founder said the actual app is being released very soon.

There’s a free version to test, and I’m paying for the premium version which allows for longer voices, customising the “translation” into my own words, and integration/webhook capability (so I can potentially build more things with it).

Check it out if it’s piqued your interest.

–

Whoops - I accidentally made a long one again.

I hope you enjoyed reading and keep the feedback, requests, comments coming - this space isn’t going to get any more ordinary any time soon and the rate of change is growing at an alarming rate.

I think the best thing we can do is simply equip ourselves with information, focus on what’s going to help us be better people and operate a better planet, and enjoy the ride.

We’ve got this! 💪

-Marshy